Predictors of cancer in patients with acute pulmonary embolism.

BACKGROUND

Acute pulmonary embolism (PE) can occur as a manifestation of an underlying cancer and be of paraneoplastic aetiology. A previously unknown cancer is sometimes diagnosed after the acute PE diagnosis. The identification of a group of patients with elevated probability of having an occult cancer underlying PE was never performed. We aimed to determine predictors of occult cancer in acute PE. Our hypothesis was that the D-dimer levels would be a predictor of cancer.

PATIENTS AND METHODS

We retrospectively analysed a cohort of patients hospitalized with acute PE.

EXCLUSION CRITERIA

<18 years, venous embolism only of veins other than pulmonary territory or when the embolism was considered chronic, and no image confirmation of acute PE. Patients were grouped according to the timing of cancer diagnosis: 1) known concomitant active cancer, 2) cancer diagnosed during acute PE admission or in the following 2 years and, 3) no known cancer during the 2-year follow-up since PE diagnosis. Predictors of concomitant cancer were determined using a logistic regression analysis. Multivariate models were built.

RESULTS

We studied 562 patients; median age was 72 years and 219 (39.0 %) were men. In 223 (39.7 %) of the patients the PE was of central arteries and 61.4 % presented with bilateral PE. PE was considered unprovoked at time of discharge in 47.7 %. Median (interquartile range) D-dimer level was 7.98 (3.30-14.99) μg/mL. A total of 126 (22.4 %) patients were in group 1, 47 in group 2 (cancer diagnosed after the diagnosis of acute PE and up to 2 years) and 389 patients were in group 3. Elevated D-dimer levels were independently associated with already known cancer. D-dimer were independent predictors of future cancer diagnosis: OR = 1.07 ((95 % CI: 1.01-1.14) per each 5 ng/mL increase; for patients with D-dimer >15.0 μg/mL the OR of future cancer was 2.10 (1.05-4.18). If only patients with unprovoked PE upon admission (n = 307) were to be considered results were similar considering D-dimer; anaemia also predicted unknown cancer [OR = 2.13 (1.08-4.16)].

CONCLUSIONS

Patients with D-dimer >15 μg/mL presented a >2-fold higher risk of being diagnosed with a cancer condition in the upcoming 2 years. D-dimer may help clinicians in identifying which patients are at higher risk of occult cancer.